No / Virtual Reality Environments, which present a safer learning and training environment, are increasingly being adopted to simulate complex systems. In parallel, distribution services have become essential following advances in telecommunications and the subsequent demand on the mobility of users. Hence, middleware enables technologies to provide such services to existing and newly-developed applications. However, distributing VRE with existing APIs still requires lots of specific development and customization.Data Distribution Service (DDS) is one of the standardized middleware for real-time applications based on a peer-to-peer architecture. It requires awareness about the types of data distributed and which is achieved by defining an information-model using an Interface Definition Language (IDL) file. Consequently, distributing VRE using DDS introduces an additional step for modelling a specific IDL file to meet each application requirements. Considering the fact that domains addressed by VRE are populated by complex data types (procedural, behavioral, etc.) then engineering a specific IDL file for each application is a complex task that requires an intervention from the computer-scientist and the domain-expert each time an application has to be distributed. The first contribution of my thesis is to provide a generic information-model which is reusable upon distributing different VRE.The novelty of our approach is based on the coupling between conceptual models (in our case we use MASCARET meta-model) and the awareness need of DDS about data to distribute, so we create generic structures within the IDL file. By this, we eliminate one step of the workflow and consequently we simplify the process of using DDS. From another side, DDS remains a low level middleware for distribution that is based on peer-to-peer architecture with no control layer in the middle. Like other classical algorithms, lots of messages should be sent over the network to synchronize the distributed environment. Moreover, we should specify by code how to detect changes in the virtual environment to send updates. Thus, the second contribution we propose is to use a generic control layer that can dynamically detect when changes occur. This layer is based onthe explicit knowledge about executed behaviors.